Thread by @jessesingal

I wrote a long critique of Gizmodo and The Markup's piece about predictive policing, which I thought exemplified many of the worst, most paternalistic trends in how liberals approach crime and policing

(For paid subscribers, but with a partial preview)

jessesingal.substack.com/p/liberal-cond…

2/ The authors argue that the main problem with PredPol, the predictive policing algorithm they're criticizing, is that black vs. white and rich vs. poor people *report* crimes at different rates. That can lead to the illusion of discrepancies. This is a very strange claim:

3/ Yes, there's some data suggesting discrepancies in the rates at which people report crime. But there's also data showing that the disparities in *actually experiencing* crime tend to be more than an order of magnitude larger. Gizmodo and The Markup ignore this entirely.

4/ Gizmodo and the Markup also treat police patrols as inherently harmful to neighborhoods where they occur, and only quote sources that agree. But this is an overwhelmingly unpopular view among black and Latino folks -- 80%+ of them *don't* want fewer cops in their area.

5/ Articles like this to be almost going out of their way to obscure the actual preferences of actual real-life human beings in high-crime areas. I think it has to do with journalists' and editors' and academics' social networks.

6/ The sentence "We did not try to determine how accurately PredPol predicted crime patterns" -- an explicit lack of curiosity about a question that could undercut this entire article -- occurs about 4,000 words in, near the end.

This is poor reporting on an important subject.